Create an account to use AI

Create a free account to unlock all features:

Ask questions about videos with AI
Use smart search (AI)
Save your favorite videos
Bookmark your favorite categories
Organize your favorite subcategories
Access your personalized favorites page
Access your conversation history
Personalize your homepage with your favorites

Create an account

Already have an account? Log in

The FASTEST introduction to Reinforcement Learning on the internet

Gonkee

Reinforcement learning is a field of machine learning concerned with how an agent should most optimally take actions in an environment. I think this field of study will be very important in the next few decades, so here I am trying to make the knowledge more accessible.

Patreon: https://www.patreon.com/Gonkee

Resources:

Reinforcement Learning: An Introduction (textbook) - http://incompleteideas.net/book/the-book-2nd.html

OpenAI Spinning Up website - https://spinningup.openai.com/en/latest/index.html

RL subreddit - https://www.reddit.com/r/reinforcementlearning/

Articles:

Deep Reinforcement Learning Doesn't Work Yet - https://www.alexirpan.com/2018/02/14/rl-hard.html

Article about monkey experiments (temporal difference & dopamine) - https://doi.org/10.1126/science.275.5306.1593

2008 article about actor-critic and the striatum - https://doi.org/10.3389/neuro.01.014.2008

Cool algorithms/papers:

MuZero - https://deepmind.google/discover/blog/muzero-mastering-go-chess-shogi-and-atari-without-rules/

DreamerV3 - https://danijar.com/project/dreamerv3/

Robot ball-in-a-cup - https://ieeexplore.ieee.org/abstract/document/5152577

Inverse Q-Learning - https://div99.github.io/IQ-Learn/

00:00 - Introduction

04:27 - Markov Decision Processes

16:37 - Grid Example + Monte Carlo

36:22 - Temporal Difference

50:54 - Deep Q Networks

58:45 - Policy Gradients

1:12:06 - Neuroscience

1:20:24 - Limitations & Future Directions

1:31:57 - Conclusion

The FASTEST introduction to Reinforcement Learning on the internet | Knoovi